BuiltForward helps contractors turn the judgment of their best people into systems the whole company can use.
AI is the enabling technology. Institutionalizing judgment is the point.
The estimator who knows which jobs to walk away from. The PM who can look at a job in month three and tell you it is going sideways before the numbers say so. The person who builds the cash forecast, and makes a dozen judgment calls inside it that nobody else in the building could make.
You can hire around those people. Most shops do.
But the decisions still come back through the same few calendars. That becomes a ceiling on growth, on how many jobs you can run at once, and on what happens when one of those people leaves or retires.
The problem is that the most valuable part of how they work usually isn't written down anywhere. Not because nobody has tried. Because the part that makes an expert good stopped feeling like knowledge to them years ago.
The standard approach to capturing institutional knowledge is to ask your best people to document it: interviews, workshops, process maps, SOPs, a binder. We think that gets the problem backwards.
People are bad at enumerating what they know. They are remarkably good at reviewing work.
Ask your best estimator to list the rules he uses and you'll get a short list followed by a lot of “it depends.” Hand him a bid that a machine put together and he'll say, “No, I would never do that,” in four seconds — and then tell you why.
That difference between what the machine did and what your expert would have done is where the valuable knowledge is hiding.
So instead of asking experts what they know, we give them work to review.
Over time, the corrections become explicit operating knowledge: rules, heuristics, thresholds, exceptions and examples that previously lived in somebody's head. That knowledge is the asset.
We don't arrive with a predetermined AI use case. During the first two weeks we walk the business department by department — Estimating, Project Management, Finance, Service and Operations — looking for work where valuable human judgment intersects with repetitive effort.
For each candidate workflow, we ask:
That last question matters. Some judgment turns out to be surprisingly consistent once you expose it. Some remains highly contextual. Sometimes repeated corrections reveal that the real problem lives somewhere upstream. Those are all useful findings.
The goal isn't to pretend every expert decision can be reduced to a rule. It's to determine how much of the expertise can be made explicit, reusable and machine-usable — and whether doing so creates enough value to matter.
We identify candidate workflows and map what lives in your accounting or project-management systems, what lives in spreadsheets, and what lives in people's heads. We score candidates on economic importance, repetitive effort, dependence on scarce people, accessibility of the information, and ability to learn through expert correction. Together we choose one, then baseline how it works today.
We build the first working version and put it in front of the expert. It attempts the work. The expert corrects it. We capture the reasoning, incorporate it, and try again — tracking whether mistakes repeat and whether new attempts need fewer interventions.
The new workflow runs alongside the real one and we compare it with the week-one baseline. Then we hand over the working workflow, the operating knowledge behind it, and a ranked map of where we think you should go next.
We choose the workflow together. We don't sell one in advance.
The right first workflow isn't necessarily the flashiest one. It's the one where better use of your people's judgment can create measurable value.
At the end of week two, we may tell you not to continue. If we don't find a workflow where we believe the remaining six weeks can create meaningful value, we stop.
You keep the workflow map, baseline and what we learned.
That outcome is built into the engagement. It's one of the possible answers, not a failure.
A lot of custom software built by an outsider becomes dead code the day that person walks out. We design against that.
The durable thing we are building is not a particular piece of software. It is the operating knowledge behind it.
The reasoning that used to live in someone's head becomes explicit: what matters, what gets flagged, what exceptions change the answer, what “good” looks like and where human judgment is still required. We keep that knowledge readable and portable rather than burying it inside a black box.
The software will change. Your systems may change. Your operating rules will evolve too. But now they can be inspected, taught, updated and carried forward rather than rediscovered every time a key person leaves.
You leave with three things: the workflow, working. The operating knowledge behind it. And a ranked map of what to do next.
That last one gets underestimated. A surprising amount of what is authoritative inside a contractor never becomes a document. Someone says it out loud, everyone understands what it means, and the company acts on it. That's exactly the knowledge we're trying to capture.
Commercial contractors and specialty subcontractors, generally $10–75 million in revenue. Big enough to have departments, systems and key people whose judgment the organization depends on. Small enough that building an internal AI or transformation team probably doesn't make sense.
If your best estimator gets eight hours a week back and there's nothing more valuable you want them doing, this is a cost rather than an investment. The strongest candidates are trying to do something: bid more work, take on larger projects, absorb an acquisition, improve margins, or prepare the business for the next generation.
This isn't an AI strategy deck. It isn't software training. It isn't a rip-and-replace of your accounting or project-management systems.
And we're not trying to replace your CPA, fractional CFO, consultants or software providers.
We work with the systems and people you already have. We're after the layer between them: the judgment your best people apply to turn information into decisions and action.
We're currently selecting a small number of contractors for the first BuiltForward engagements. These are fixed-fee, eight-week engagements offered at a substantially reduced design-partner rate.
In exchange, we ask for unusually close access to the work, candid feedback about what does and doesn't work, and permission to incorporate anonymized structural learnings into the BuiltForward method.
Your company name, financial information, customers, projects and proprietary information remain confidential. The workflows and operating knowledge we build specifically for your company are yours.
BuiltForward was founded by Jason Jacobs, previously founder of Runkeeper, which was acquired by ASICS, and co-founder of MCJ.
It grew out of conversations with contractor owners, CFOs, estimators, project leaders, sureties, consultants and software companies about a deceptively simple question:
As AI becomes capable of doing more of the work, how do we make sure it learns how your best people actually do it?
Every week on the BuiltForward podcast, Jason has a long conversation with someone who knows something he doesn't — contractor owners, the people who finance and bond them, and the companies building software for them. It's an open notebook, not a highlight reel. Where the thinking on this page came from is mostly there.
Watch on YouTube — or listen on Apple Podcasts and Spotify
Guests come on to talk about their own work. They are not clients, and appearing here isn't an endorsement of BuiltForward.
Tell us what runs through your best few people. We'll tell you honestly whether we think there's a workflow here worth eight weeks.
jason@builtforward.aiNot there yet? Start with the podcast. It's the same thinking, at no commitment, and it will tell you fairly quickly whether we're asking useful questions.